1. |
|
|
2. |
|
|
3. |
- Kofler, Florian, et al.
(författare)
-
Approaching Peak Ground Truth
- 2023
-
Ingår i: Proceedings - International Symposium on Biomedical Imaging. - 1945-7928 .- 1945-8452. - 9781665473583
-
Konferensbidrag (refereegranskat)abstract
- Machine learning models are typically evaluated by computing similarity with reference annotations and trained by maximizing similarity with such. Especially in the biomedical domain, annotations are subjective and suffer from low inter-and intra-rater reliability. Since annotations only reflect one interpretation of the real world, this can lead to sub-optimal predictions even though the model achieves high similarity scores. Here, the theoretical concept of Peak Ground Truth (PGT) is introduced. PGT marks the point beyond which an increase in similarity with the reference annotation stops translating to better Real World Model Performance (RWMP). Additionally, a quantitative technique to approximate PGT by computing inter- and intra-rater reliability is proposed. Finally, four categories of PGT-aware strategies to evaluate and improve model performance are reviewed.
|
|
4. |
|
|
5. |
- McKee, Andrew, et al.
(författare)
-
Central and convolution Herz-Schur multipliers
- 2022
-
Ingår i: New York Journal of Mathematics. - 1076-9803 .- 1076-9803. ; 28, s. 1-43
-
Tidskriftsartikel (refereegranskat)abstract
- In this paper we obtain descriptions of central operator-valued Schur and Herz-Schur multipliers, akin to a classical characterisation due to Grothendieck, that reveals a close link between central (linear) multipliers and bilinear multipliers into the trace class. Restricting to dynamical systems where a locally compact group acts on itself by translation, we identify their convolution multipliers as the right completely bounded multipliers, in the sense of Junge-Neufang-Ruan, of a canonical quantum group associated with the underlying group. We provide characterisations of contractive idempotent operator-valued Schur and Herz-Schur multipliers. Exploiting the link between Herz-Schur multipliers and multipliers on transformation groupoids, we provide a combinatorial characterisation of groupoid multipliers that are contractive and idempotent.
|
|
6. |
|
|
7. |
- Pigeot, Sébastien, et al.
(författare)
-
Orthotopic Bone Formation by Streamlined Engineering and Devitalization of Human Hypertrophic Cartilage
- 2020
-
Ingår i: International Journal of Molecular Sciences. - : MDPI AG. - 1422-0067. ; 21:19
-
Tidskriftsartikel (refereegranskat)abstract
- Most bones of the human body form and heal through endochondral ossification, whereby hypertrophic cartilage (HyC) is formed and subsequently remodeled into bone. We previously demonstrated that HyC can be engineered from human mesenchymal stromal cells (hMSC), and subsequently devitalized by apoptosis induction. The resulting extracellular matrix (ECM) tissue retained osteoinductive properties, leading to ectopic bone formation. In this study, we aimed at engineering and devitalizing upscaled quantities of HyC ECM within a perfusion bioreactor, followed by in vivo assessment in an orthotopic bone repair model. We hypothesized that the devitalized HyC ECM would outperform a clinical product currently used for bone reconstructive surgery. Human MSC were genetically engineered with a gene cassette enabling apoptosis induction upon addition of an adjuvant. Engineered hMSC were seeded, differentiated, and devitalized within a perfusion bioreactor. The resulting HyC ECM was subsequently implanted in a 10-mm rabbit calvarial defect model, with processed human bone (Maxgraft®) as control. Human MSC cultured in the perfusion bioreactor generated a homogenous HyC ECM and were efficiently induced towards apoptosis. Following six weeks of in vivo implantation, microcomputed tomography and histological analyses of the defects revealed an increased bone formation in the defects filled with HyC ECM as compared to Maxgraft®. This work demonstrates the suitability of engineered devitalized HyC ECM as a bone substitute material, with a performance superior to a state-of-the-art commercial graft. Streamlined generation of the devitalized tissue transplant within a perfusion bioreactor is relevant towards standardized and automated manufacturing of a clinical product.
|
|
8. |
- Selvaraju, Ram K., et al.
(författare)
-
In Vivo Imaging of the Glucagonlike Peptide 1 Receptor in the Pancreas with Ga-68-Labeled DO3A-Exendin-4
- 2013
-
Ingår i: Journal of Nuclear Medicine. - : Society of Nuclear Medicine. - 0161-5505 .- 1535-5667 .- 2159-662X. ; 54:8, s. 1458-1463
-
Tidskriftsartikel (refereegranskat)abstract
- The glucagonlike peptide 1 receptor (GLP-1R) is mainly expressed on beta-cells in the Wets of Langerhans and is therefore an attractive target for imaging of the beta-cell mass. In the present study, Ga-68-labeled exendin-4 was evaluated for PET imaging and quantification of GLP-1R in the pancreas. Methods: Dose escalation studies of Ga-68-labeled 1,4,7-tris(carboxymethylaza)cyclododecane-10-azaacetyl (DO3A)-exendin-4 were performed in rats (organ distribution) and cynomolgus monkeys (PET/CT imaging) to determine the GLP-1R-specific tissue uptake in vivo. Pancreatic uptake (as determined by organ distribution) in healthy rats was compared with that in diabetic rats. GLP-1R occupancy in the cynomolgus pancreas was quantified with a 1-tissue-compartment model. Results: In rodents, uptake in the pancreas was decreased from the baseline by up to 90% (P < 0.0001) by coadministration of DO3A-exendin-4 at 100 mu g/kg. Pancreatic uptake in diabetic animals was decreased by more than 80% (P < 0.001) compared with that in healthy controls, as measured by organ distribution. GLP-1R occupancy in the cynomolgus pancreas after coinjection of DO3A-exendin-4 at 0.15-20 mu g/kg ranged from 49% to 97%, as estimated by compartment modeling. Conclusion: These results strongly support the notion that Ga-68-DO3A-exendin-4 uptake in the pancreas is mediated by specific receptor binding. In addition, pancreatic uptake was decreased by selective destruction of beta-cells. This result suggests that GLP-1R can be quantified in vivo, which has major implications for the prospect of imaging of native beta-cells.
|
|
9. |
- Selvaraju, Ram Kumar, et al.
(författare)
-
Pre-clinical evaluation of [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4 for imaging of insulinoma
- 2014
-
Ingår i: Nuclear Medicine and Biology. - : Elsevier BV. - 0969-8051 .- 1872-9614. ; 41:6, s. 471-476
-
Tidskriftsartikel (refereegranskat)abstract
- INTRODUCTION: Insulinoma is the most common form of pancreatic endocrine tumors responsible for hyperinsulinism in adults. These tumors overexpress glucagon like peptide-1 (GLP-1) receptor, and biologically stable GLP-1 analogs have therefore been proposed as potential imaging agents. Here, we evaluate the potential of a positron emission tomography (PET) tracer, [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4, for imaging and quantification of GLP-1 receptors (GLP-1R) in insulinoma.METHODS: [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4 was evaluated for binding to GLP-1R by in vitro autoradiography binding studies in INS-1 tumor from xenografts. In vivo biodistribution was investigated in healthy control mice, INS-1 xenografted and PANC1 xenografted immunodeficient mice at two different doses of peptide: 2.5μg/kg (baseline) and 100μg/kg (block). In vivo imaging of [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4 in xenografted mice was evaluated by small animal PET/CT using a direct comparison with the clinically established insulinoma marker [(11)C]5-hydroxy-tryptophan ([(11)C]5-HTP).RESULTS: GLP-1 receptor density could be quantified in INS-1 tumor biopsies. [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4 showed significant uptake (p≤0.05) in GLP1-R positive tissues such as INS-1 tumor, lungs and pancreas upon comparison between baseline and blocking studies. In vivo imaging showed concordant results with higher tumor-to-muscle ratio in INS-1 xenografted mice compared with [(11)C]5-HTP.CONCLUSION: [(68)Ga]Ga-DO3A-VS-Cys(40)-Exendin-4 has high affinity and specificity for GLP-1R expressed on insulinoma in vitro and in vivo.
|
|
10. |
|
|